/
model.py
187 lines (143 loc) · 6.6 KB
/
model.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
import torch
import torch.nn as nn
import torch.nn.functional as F
from attention import AttentionConv
from attention_augmented_conv import AugmentedConv
#TODO Make width not increase with # groups
class Bottleneck(nn.Module):
expansion = 4
def __init__(self, in_channels, out_channels, stride=1, base_width=64, args=None):
super(Bottleneck, self).__init__()
self.stride = stride
groups = args.groups # Number of attention heads
'''
# TODO : Doubt in width, when base_width != 64?
width = int(out_channels * (base_width / 64.))\
if args.attention_conv\
else int(out_channels * (base_width / 64.)) * groups
'''
width = out_channels
additional_args = {'groups':groups, 'R':args.R, 'z_init':args.z_init, 'adaptive_span':args.adaptive_span} \
if args.all_attention else {'bias': False}
kernel_size = args.attention_kernel if args.all_attention else 3
padding = int((kernel_size - 1) / 2)
layer = None
if args.attention_conv:
layer = AugmentedConv(width, width, kernel_size, args.dk, args.dv, groups, shape=width)
elif args.all_attention:
layer = AttentionConv(width, width, kernel_size=kernel_size, padding=padding, **additional_args)
else:
layer = nn.Conv2d(width, width, kernel_size=kernel_size, padding=padding, **additional_args)
self.conv1 = nn.Sequential(
nn.Conv2d(in_channels, width, kernel_size=1, bias=False),
nn.BatchNorm2d(width),
nn.ReLU(),
)
self.conv2 = nn.Sequential(
layer,
nn.BatchNorm2d(width),
nn.ReLU(),
)
self.conv3 = nn.Sequential(
nn.Conv2d(width, self.expansion * out_channels, kernel_size=1, bias=False),
nn.BatchNorm2d(self.expansion * out_channels),
)
self.shortcut = nn.Sequential()
if stride != 1 or in_channels != self.expansion * out_channels:
self.shortcut = nn.Sequential(
nn.Conv2d(in_channels, self.expansion * out_channels, kernel_size=1, stride=stride, bias=False),
nn.BatchNorm2d(self.expansion * out_channels)
)
def forward(self, x):
out = self.conv1(x)
out = self.conv2(out)
out = self.conv3(out)
if self.stride >= 2:
out = F.avg_pool2d(out, (self.stride, self.stride))
out += self.shortcut(x)
out = F.relu(out)
return out
class Model(nn.Module):
def __init__(self, block, num_blocks, num_classes=1000, args=None):
super(Model, self).__init__()
divider = 2 #if args.small_version else 1
layer_channels = None #These two sets of channels give approximately equal #params between all_attention and all_conv
if args.all_attention:
#layer_channels = [64,128,128,256,256]
layer_channels = [32, 64, 128] if args.smallest_version else [64//divider, 128//divider, 256//divider, 512//divider] # [96, 128, 128, 256]
else:
layer_channels = [32, 64, 128] if args.smallest_version\
else [64//divider, 128//divider, 256//divider, 512//divider]
self.args = args
self.in_places = 64 if args.dataset == 'TinyImageNet' else 32
self.all_attention = args.all_attention
self.attention_kernel = args.attention_kernel
self.init = nn.Sequential(
nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False),
nn.BatchNorm2d(64),
nn.ReLU(),
nn.MaxPool2d(kernel_size=3, stride=2, padding=1),
) if args.dataset == 'TinyImageNet' else nn.Sequential(
nn.Conv2d(3, 64 // divider, kernel_size=3, stride=1, padding=1, bias=False),
nn.BatchNorm2d(64 // divider),
nn.ReLU()
)
self.layers = nn.ModuleList()
strides = [2]*3 if args.smallest_version else [1] + [2]*3
for i in range(len(layer_channels)):
self.layers.append(self._make_layer(block, layer_channels[i], num_blocks[i], stride=strides[i]))
self.dense = nn.Linear(layer_channels[-1] * block.expansion, num_classes)
def _make_layer(self, block, planes, num_blocks, stride):
strides = [stride] + [1] * (num_blocks - 1)
layers = []
for stride in strides:
layers.append(block(self.in_places, planes, stride, args=self.args)) #in_places is #input_channels
self.in_places = planes * block.expansion
return nn.Sequential(*layers)
def get_span_l1(self, args):
num_abs_spans = 0
if args.all_attention:
for l in self.layers:
for l2 in l:
sum_layer = l2.conv2[0].adaptive_mask.current_val.abs().sum()
num_abs_spans += sum_layer
return num_abs_spans
def clamp_span(self):
for l in self.layers:
for l2 in l:
l2.conv2[0].adaptive_mask.clamp_param()
def forward(self, x):
# TODO(Joe): See if there is some other modification we can make so we don't need to have different pooling kernels at the end of the model
pooling_kernel_size = 2 if self.args.dataset == 'TinyImageNet' else 4
out = self.init(x)
for layer in self.layers:
out = layer(out)
out = F.avg_pool2d(out, pooling_kernel_size)
out = out.view(out.size(0), -1)
out = self.dense(out)
return out
def ResNet26(num_classes=1000, args=None):
if args.smallest_version:
num_blocks = [1]*3
elif args.small_version:
#Decided to use same architecture for both all conv and all attention for better comparison
num_blocks = [1]*4 #[1, 2, 2, 1] if args.all_attention else [1]*4
else:
num_blocks = [1,3,4,1] #Now all attention is 3.02M and CNN is 3.09 M params
#num_blocks = [1, 2, 4, 1]
return Model(Bottleneck, num_blocks, num_classes=num_classes, args=args)
def ResNet38(num_classes=1000, all_attention=False):
return Model(Bottleneck, [2, 3, 5, 2], num_classes=num_classes)
def ResNet50(num_classes=1000, all_attention=False):
return Model(Bottleneck, [3, 4, 6, 3], num_classes=num_classes)
def get_model_parameters(model):
total_parameters = 0
for layer in list(model.parameters()):
layer_parameter = 1
for l in list(layer.size()):
layer_parameter *= l
total_parameters += layer_parameter
return total_parameters
# temp = torch.randn((2, 3, 224, 224))
# model = ResNet38(num_classes=1000)
# print(get_model_parameters(model))